A field experiment was conducted during the period from October 2001 to April 2002 to study the effect of earthing up and level of irrigation on yield and quality seed production of onion (cv. Taherpuri) at Rajshahi University campus, Rajshahi, Bangladesh. Two different factors were considered, factor (A): Earthing up (E1), without earthing up (E0) and factor (B): irrigation level (5 levels) viz. irrigation start at 1 st time and when required (I 4 ), irrigation start at 40 DAP and when required (I 3 ), irrigation start at 60 DAP and when required (I2), irrigation start at 80 DAP and when required (I1), no irrigation or control (I 0 ). The standard dose of cowdung and NPK were applied according to HRDP (1995). The factorial experiment was laid out following at Split Plot Design (SPD) with three replications. The result revealed that bulb emergence, plant height number of leaves, length of scape, tillers, seeded fruits, fruits set, days to blooming, seed yield, 1000-seed weight and germination percentage were significantly influenced by different treatments. The highest seed yield (405.97 kg/ha) was found from earthing up with irrigation start at first time and when required (E 1 I 4 ) followed by E 1 I 3 (347.37 kg/ha), E0I4 (330.73 kg/ha) compared with other treatments. The results suggested that earthing up with 3-4 times irrigation is more effective for onion seed production in Bangladesh.

Rainfall is the main source of irrigation water in the northwest part of Bangladesh where the inhabitants derive their income primarily from farming. Stochastic rainfall models were concerned with the occurrence of wet day and depth of rainfall. The first order Markov chain model was used to generate the sequence of rainfall occurrence using the method of transitional probability matrices, while daily rainfall amount was generated using a gamma distribution. The model parameters were estimated from historical rainfall records. The shape and scale parameters were estimated by method of moments and hence it became possible to find the parameter values at the study area and then to generate synthetic sequences according to the gamma distribution. The parameters necessary for the whole generation include the means, variance or standard deviation and conditional probabilities of wet and dry days. Results obtained showed that the model could be used to generate rainfall data satisfactorily.

No abstract